The Real Eisman Playbook Sep 18, 2026 · recorded Sep 17, 2026 25m 13m saved
About $500 billion of debt has been raised this year to fund artificial intelligence, and this episode's argument is that the money is being taken out of the Treasury market.
The consensus explanation for a 10-year yield above 5% is the Iran war, the oil price and inflation. This show agrees those matter and says something else is running underneath them: investors buying AI paper instead of government paper.
"This year, approximately 500 billion in AI related debt has been raised. That's a lot of long-term debt and it is creating a crowding out effect."
The Real Eisman Playbook is written and presented by one of the investors behind the Big Short trade, who has been publicly short FICO through a 43% decline this year. This weekly wrap covered five subjects, and the last of them is a theory about why the two largest artificial intelligence companies started warning that their own product might kill everyone.
The full episode is covered here so you can skip it. 25 minutes of audio, 12 minutes of reading.
Here are the 12 arguments that matter.
Key Takeaways
5% on the 10-year is treated as the line the market cannot hold above, after an earlier call of 4.5% that proved wrong
The $500B of AI debt raised this year is crowding out Treasuries, which is the part of the yield story nobody is pricing
Scott Bessent's buyback of long-dated Treasuries, $4B raised to $6B, worked for one day
FICO raised prices 1,600% over five years, and the regulator of Fannie Mae and Freddie Mac has made that a stock problem
VantageScore has 10% of newly securitized mortgages, which is the wedge in the monopoly
The claim that Anthropic and OpenAI want to slow down is called false on its face: OpenAI is $300B of Oracle's backlog
The doomsday campaign is read as an attempt to invite regulation and build an AI duopoly
Only staples and healthcare, 14% of the S&P 500, are uncorrelated to AI
Shorting against the box is offered as the way to cut technology exposure without paying capital gains tax
1. Oil, Rates and the Fed
The war in Iran continues with no letup and both sides are doing damage, which took oil close to $110 before it pulled back. The inflation scare that followed pushed the 10-year Treasury yield above 5% for a time.
The level he got wrong, and the one that holds
Now, for quite some time, I have been saying that the market will not be able to stomach some level of long-term rates. Originally, I thought that level was 4.5%. And that was wrong.
The host
The Federal Reserve raised rates by 25 basis points this week, to a range of 3.75% to 4%, and penciled in another increase later this year, in defiance of the president's calls for cuts. The market's objection on Wednesday was to the signal of more rather than to the rise itself. On Thursday long rates fell back below 5% as the market assumed higher short rates will slow the economy, and equities recovered.
Where the line sits now
For now, 5% on the 10-year does seem to be the demarcation line.
The host
2. $500B of AI Debt
The war, the oil price and inflation all contributed to higher long-term yields. The episode's addition to that list is supply of a different kind of bond.
The crowding-out claim
This year, approximately 500 billion in AI related debt has been raised. That's a lot of long-term debt and it is creating a crowding out effect.
The host
Which shows up as a choice between two long bonds
In other words, some investors would rather buy AI long-term debt than long-term US treasuries.
The host
3. Bessent Needs a Bazooka
With US debt at $40 trillion, every rise in long-term yields adds to the deficit, which is why the Treasury Secretary wants those yields lower.
The position he is in
It's a perfect storm for Treasury Secretary Scott Bessent, and I don't envy him his position.
The host
The response was a program to buy $4 billion of long-dated Treasuries, later increased to $6 billion. It held for a single day before yields resumed rising.
What the show says is needed instead
He needs a much bigger bazooka or an alternative buyer.
The host
Bessent is not expected to disappear; a new plan is expected, though what it contains is not something the episode would predict.
4. Still Short FICO
The short case on FICO is a pricing case, and the position has been held for a while.
The thesis in one line
My thesis was that the company got greedy and raised prices 1,600%.
The host
That is over five years, in a monopoly on mortgage credit scoring that the show says has been abused. The reason it has become tradeable is that Bill Pulte, who runs the Federal Housing Finance Agency and therefore regulates Fannie Mae and Freddie Mac, has been attacking the company relentlessly. A pilot program last year let 21 lenders use VantageScore instead; a few weeks ago Pulte said on social media that the program was good for all lenders, and criticized the credit bureaus for price gouging as well.
What that has done to the three stocks
Year-to date, FICO is down 43% and Equifax and TransUnion are down 25% and 15% respectively.
The host
Equifax has fallen further than TransUnion because a larger share of its profit comes from mortgage scoring.
The call from here
I remain short FICO and think that its monopoly in mortgage scoring is going to break.
The host
VantageScore had 10% of new mortgage loans securitized on the most recent data, a share the show expects to go much higher.
5. Pulte and the Insurers
Pulte's next target is the mortgage insurance sector: MGIC, Essent and Radian. The episode explained the product before the argument.
Who the policy actually protects
You, the borrower, pay for the mortgage insurance, but it does not benefit you. Mortgage insurance is designed to protect the lender.
The host
If a home is foreclosed and sells for less than the loan, the insurer pays the lender the difference. Pulte's case is that the industry should be more consumer-friendly and should do more to tell borrowers when the insurance is no longer needed.
6. The Clarity Act Blocked
The crypto industry has spent years and hundreds of millions of dollars on the Clarity Act, a bill designed to give digital assets a clear rulebook in the US. A few weeks ago it looked likely to pass, which the show reads as the explanation for the rally in cryptocurrencies and in Circle, the stablecoin company.
Then the Senate stopped it
This is a major blow for the crypto industry and it is unclear where the industry goes from here.
The host
Democrats cited concerns over ethics provisions and a few Republicans joined them.
7. The Doomsday Posts
The sequence started with Jacob Coxon, a 27-year-old artificial intelligence researcher.
The post that started it
He resigned from Anthropic and put out a social media post claiming that AI labs are racing toward self-improving super intelligence without proper controls.
The host
Coxon also claimed the people building the technology privately believe it could kill everyone by the end of the decade. The post drew millions of views. Dario Amodei, Anthropic's chief executive, then called for the industry to slow the pace of capability improvements, and gave CBS News a line the episode quoted.
What Amodei told CBS News
For too long, the industry lied or downplayed the real risks of AI.
Dario Amodei, quoted on the show
Sam Altman of OpenAI said he agreed the industry needs to slow down. The show's objection is what the two of them are not discussing: teenage suicides connected to interaction with these products, and illegal hacking of other companies, both of which it describes as solvable with adequate oversight by the same managements.
8. The Whistling Hacker
For a precedent, the episode borrowed a story from Ed Zitron's newsletter, credited by name, about the hacker Kevin Mitnick, convicted in 1988, released, reconvicted in 1995 and given five years with eight months in solitary confinement. Prosecutors persuaded a judge that Mitnick could start a nuclear war by whistling into a prison pay phone to reach a NORAD modem.
Why that story is the model
But it was new technology that was poorly understood by most people. And these lay people could be convinced of nonsense.
The host
The other precedent is nearer. Artificial intelligence executives spent a year predicting mass layoffs across the economy, which has not happened, and have stopped talking about it.
The pattern that leaves
So the AI industry has a tendency to make bold and terrible predictions that just don't come true.
The host
Reporters, politicians and other decision-makers accept each claim, on this account, because the subject is new enough that they have nothing to test it against.
9. The Subterfuge Theory
The show's own explanation is that the slowdown talk is a trick, and the first piece of evidence is the business itself.
The claim cannot be true on its face
The entire future of Anthropic and OpenAI depends on there not being any slowdown.
The host
Both have hundreds of billions of dollars of commitments to hyperscalers.
The single number that shows the size of it
OpenAI alone represents 300 billion of Oracle's 600 billion plus backlog.
The host
Anthropic is expected to go public this year, possibly within a month or two, which is the other reason the show says a slowdown cannot be the message.
What a road show could not say
A slowdown contradicts their entire growth narrative.
The host
The version of that argument the episode liked best
It's certainly some crazy 4D chess to say that there's whatever a 10% chance of annihilating humanity, but by the way, how much allocation would you like in our IPO?
Elon Musk, quoted on the show
10. A Price War Coming
Underneath the theory is a claim about the economics of selling a model.
Why the business is getting harder
Business is potentially slowing because token maxing is ending and open weight models keep taking market share. There are no pricing moats in this business.
The host
Today's best model is tomorrow's second-best and cheaper, while the cost of building data centers and the cost of capital both rise. A business with no moat and a price war coming needs one from somewhere.
Where the show says they are looking for it
So they are fomenting hysteria thereby inviting the federal government to regulate the industry.
The host
Regulation, on this reading, would create the moats and leave a duopoly. The gambit is also judged to have failed: the president said this week he has no interest in regulating artificial intelligence, so the episode expects the doomsday narrative to disappear quickly. What it leaves behind does not.
The damage it counts as already done
AI won't cause extinction, but these two CEOs are creating massive damage.
The host
Data center regulation is now the centerpiece of many local elections, with voters who believe the building next door could kill them. Lying to investors is described as old-school Wall Street fraud; lying to politicians is described as capable of disrupting the whole economy.
The test that comes next
It's unclear to me how to keep a shell game going in an IPO process that requires transparency.
The host
11. Only 14% Is Uncorrelated
A viewer asked how an individual investor diversifies away from artificial intelligence without leaving equities. The answer started by widening the problem: utilities are correlated because they power the data centers, industrials because they build them, and large banks, investment banks and alternative managers because they finance them.
What is left
There are 11 sectors of the S&P 500. And as far as I can tell, the only sectors that are uncorrelated are staples and healthcare, which combined are only 14% of the S&P 500.
The host
A healthcare or staples exchange-traded fund is the blunt version. The finer version is that almost every sector contains an uncorrelated subsector, with property and casualty insurance inside financials the example given, uncorrelated to artificial intelligence and to the economy. Three funds were named: the Franklin US Low Volatility High Dividend Index ETF, the Invesco S&P 500 Low Volatility ETF and the Invesco KBW Property & Casualty Insurance ETF.
12. Shorting Against the Box
The second part of that answer was for a holder sitting on large unrealized gains in technology who does not want to sell.
The problem stated with a position
Let's say I own 100 shares of Nvidia and I've owned it for years. So, I have massive embedded gains.
The host
Selling means paying 20% federal long-term capital gains tax plus the state rate. Shorting 50 or 100 shares of the same stock instead is what the show means by shorting against the box.
What that buys
By doing so, I have now reduced my Nvidia risk and I can wait until I think things will get better and then I can undo the short.
The host
The practical note attached: an account has to be a margin account to short, which takes signing a document.
Bonus Insights
A viewer asked whether the market is underpricing the investment income property and casualty insurers earn on their float as bond books roll into higher yields. The episode agreed with the mechanism and disagreed about what it is worth.
What the question misses
What I think he might be missing is that generally investors in P&C stocks don't care all that much when P&C companies make more money on the float.
The host
Pricing is what those investors watch, so rising rates alongside falling pricing still puts the stocks under pressure.
Another viewer pushed back on a previous interview with Wolfgang Munchau, author of "Kaput," arguing it concentrated on the UK, Germany, France and Italy while ignoring Sweden, with debt at 36% of GDP, and Poland's recent growth. The episode accepted the criticism as fair and then restated the case.
Why the big four decide the answer
These European countries have small tech sectors and since tech is what drives growth, European growth is generally slow.
The host
Europe was also described as having a richer welfare system, being overregulated and being almost allergic to risk.
The bottom line across the whole episode is that the AI build-out is now the marginal buyer of long-term credit and therefore part of the reason Treasury yields are where they are, and that the same build-out's two loudest voices are talking about extinction because the alternative is talking about their margins.
Products, Companies & Tools Mentioned
FICO (The short: a 1,600% price increase over five years in mortgage credit scoring, with the stock down 43% year to date)
Equifax and TransUnion (Down 25% and 15% year to date; Equifax has more mortgage-scoring profit at risk)
VantageScore (The FICO alternative, at 10% of newly securitized mortgages and expected to take more)
MGIC, Essent and Radian (The mortgage insurers the FHFA director has turned to next)
Anthropic and OpenAI (The two companies whose slowdown talk the episode treats as a bid for regulatory moats)
Oracle (Its backlog is the evidence: $300B of $600B-plus comes from OpenAI alone)
Circle (The stablecoin company whose share price the show ties to the odds on the Clarity Act)
Nvidia (The worked example for shorting against the box rather than realizing a gain)
Invesco S&P 500 Low Volatility ETF, Invesco KBW Property & Casualty Insurance ETF and the Franklin US Low Volatility High Dividend Index ETF (The three funds named for diversifying away from the AI trade)
Books & Resources Mentioned
Kaput – Wolfgang Munchau (Introduced on the show as "Kaput, the end of the German economic miracle"; the subject of a previous interview and of a viewer's challenge on Sweden and Poland)
Ed Zitron's newsletter (Where the Kevin Mitnick story came from, credited on air)
If this was worth your time, send it to someone closer to the industry than you are.
Get the latest market chatter as it happens: